Microfluidic Device for Electric Field-Driven Single-Cell Capture and

S. Yen , Jiasi Wang , Shihan Xu , Min Li , Amy L. Paguirigan , Jordan L. Smith , Jerald .... Erik S. Douglas , Sonny C. Hsiao , Hiroaki Onoe , Car...
2 downloads 0 Views 282KB Size
Download PDF
Anal. Chem. 2005, 77, 6935-6941

Microfluidic Device for Electric Field-Driven Single-Cell Capture and Activation Nicholas M. Toriello,† Erik S. Douglas,† and Richard A. Mathies*

Department of Chemistry and UC Berkeley/UCSF Joint Graduate Group in Bioengineering, University of California, Berkeley, California 94720

A microchip that performs directed capture and chemical activation of surface-modified single cells has been developed. The cell capture system is comprised of interdigitated gold electrodes microfabricated on a glass substrate within PDMS channels. The cell surface is labeled with thiol functional groups using endogenous RGD receptors, and adhesion to exposed gold pads on the electrodes is directed by applying a driving electric potential. Multiple cell types can thus be sequentially and selectively captured on desired electrodes. Single-cell capture efficiency is optimized by varying the duration of field application. Maximum single-cell capture is attained for the 10-min trial, with 63 ( 9% (n ) 30) of the electrode pad rows having a single cell. In activation studies, single M1WT3 CHO cells loaded with the calciumsensitive dye fluo-4 AM were captured; exposure to the muscarinic agonist carbachol increased the fluorescence to 220 ( 74% (n ) 79) of the original intensity. These results demonstrate the ability to direct the adhesion of selected living single cells on electrodes in a microfluidic device and to analyze their response to chemical stimuli. The ability to selectively position and interface with single cells from different populations offers many opportunities for studying cell-cell signaling,1 genetic heterogeneity,2 and heterotypic biological systems.3 While significant progress has been made in monitoring bulk electrical, mechanical, chemical, and genetic changes in cell populations, a fundamental understanding of cellular interactions and stochastic effects at the single-cell level has yet to be fully realized.4,5 Measurements from the elements of a heterotypic array of cells reveal unique individual behavior while maintaining their natural context. Ideally a single-cell analysis system would allow for the immobilization, stimulation, and recording (electrical, mechanical, biochemical, or optical) from multiple cell types in a fast, high-throughput, and highly parallel manner. In addition to addressing the opportunities mentioned * To whom correspondence should be addressed. Phone: (510) 642-4192. Fax: (510) 642-3599. E-mail: [email protected]. † These authors contributed equally to this work. (1) Folch, A.; Toner, M. Annu. Rev. Biomed. Eng. 2000, 2, 227-256. (2) Hahn, S.; Zhong, X. Y.; Troeger, C.; Burgemeister, R.; Gloning, K.; Holzgreve, W. Cell. Mol. Life Sci. 2000, 57, 96-105. (3) Rubin, M. A. Science 2002, 296, 1329-1330. (4) Andersson, H.; van den Berg, A. Curr. Opin. Biotechnol. 2004, 15, 44-49. (5) Brehm-Stecher, B. F.; Johnson, E. A. Microbiol. Mol. Biol. Rev. 2004, 68, 538-559. 10.1021/ac051032d CCC: $30.25 Published on Web 10/04/2005

© 2005 American Chemical Society

above, this system may be the precursor to a general bioelectronic interface. Microfabricated devices provide an excellent platform with which to explore changes and interactions at the single-cell level due to their comparable size scales (1-100 µm), patternable interfaces, and the opportunity for parallel analysis.6 Mechanical cell trapping,7-9 cell sorting,10 optical tweezers,11 dielectrophoresis,12 and electroaddressable array13 microdevices offer the ability to quickly isolate and probe individual cells. Novel surface engineering approaches have been developed for the micropatterning of single-cell attachment islands14,15 and for the co-culture of multiple cell types on a single substrate.16-19 While previous work has demonstrated cell manipulation, patterning, and coculture of multiple cell types on a single substrate, an integrated microdevice with the ability to position and interface with single cells from different populations has yet to be developed. Here we demonstrate a novel combination of cell surface modification and electric field-directed adhesion and its use for the rapid capture and chemical activation of living single cells in a microchip. Our approach is to label the cell surface with thiol functional groups using endogenous receptors to the cell adhesion peptide sequence RGD.20,21 The labeled cells are electrophoreti(6) Andersson, H.; van den Berg, A. Sens. Actuators, B 2003, 92, 315-325. (7) Wheeler, A. R.; Throndset, W. R.; Whelan, R. J.; Leach, A. M.; Zare, R. N.; Liao, Y. H.; Farrell, K.; Manger, I. D.; Daridon, A. Anal. Chem. 2003, 75, 3581-3586. (8) Wu, H. K.; Wheeler, A.; Zare, R. N. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 12809-12813. (9) Huang, W. H.; Cheng, W.; Zhang, Z.; Pang, D. W.; Wang, Z. L.; Cheng, J. K.; Cui, D. F. Anal. Chem. 2004, 76, 483-488. (10) Fu, A. Y.; Spence, C.; Scherer, A.; Arnold, F. H.; Quake, S. R. Nat. Biotechnol. 1999, 17, 1109-1111. (11) Munce, N. R.; Li, J. Z.; Herman, P. R.; Lilge, L. Anal. Chem. 2004, 76, 4983-4989. (12) Voldman, J.; Gray, M. L.; Toner, M.; Schmidt, M. A. Anal. Chem. 2002, 74, 3984-3990. (13) Ozkan, M.; Pisanic, T.; Scheel, J.; Barlow, C.; Esener, S.; Bhatia, S. N. Langmuir 2003, 19, 1532-1538. (14) Chen, C. S.; Mrksich, M.; Huang, S.; Whitesides, G. M.; Ingber, D. E. Science 1997, 276, 1425-1428. (15) Veiseh, M.; Wickes, B. T.; Castner, D. G.; Zhang, M. Q. Biomaterials 2004, 25, 3315-3324. (16) Bhatia, S. N.; Balis, U. J.; Yarmush, M. L.; Toner, M. FASEB J. 1999, 13, 1883-1900. (17) Yousaf, M. N.; Houseman, B. T.; Mrksich, M. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 5992-5996. (18) Takano, H.; Sul, J. Y.; Mazzanti, M. L.; Doyle, R. T.; Haydon, P. G.; Porter, M. D. Anal. Chem. 2002, 74, 4640-4646. (19) Co, C. C.; Wang, Y. C.; Ho, C. C. J. Am. Chem. Soc. 2005, 127, 1598-1599. (20) Pierschbacher, M. D.; Ruoslahti, E. Proc. Natl. Acad. Sci. U.S.A. 1984, 81, 5985-5988.

Analytical Chemistry, Vol. 77, No. 21, November 1, 2005 6935

cally directed to selected gold electrodes, due to their intrinsic negative surface charge.22 Once captured, the single cells are activated with an agonist to a membrane-bound receptor, and the response is monitored optically with a fluorescent probe. Multiple cell types are sequentially and selectively captured on neighboring electrodes by changing the field direction. Such directed capture of multiple viable cell types in a microfluidic device presents a new paradigm for analyzing the activity and interactions of single cells. EXPERIMENTAL SECTION Microchip Fabrication. Gold electrodes are fabricated on cleaned 4-in.-diameter borofloat glass wafers utilizing a liftoff protocol. Bare glass wafers are treated with hexamethyldisilazane for 5 min to promote photoresist adhesion and Shipley 1818 photoresist (PR) is spun at 2500 rpm for 30 s and cured at 120 °C for 90 s. The interdigitated 40-µm-wide electrodes are patterned in a Karl Suss MA6 Mask Aligner through a chrome mask. Exposed positive photoresist is removed in 1:1 diluted Microposit Developer. A 20-nm seed layer of chromium and 100-nm gold film are evaporated onto the patterned glass substrate. The gold electrodes are defined by placing the glass substrates in photoresist stripping solution, PRS-3000 (Mallinckrodt Baker, Phillipsburg, NJ), lifting off chrome and gold from unexposed regions. An oxide layer was deposited and patterned, leaving three 16µm2 single-cell adhesion pads exposed on each gold electrode. A 24-nm layer of SiO2 is deposited on the wafer surface using a Randex Sputtering System. A masking layer of PR is then lithographically patterned on the surface, with the 16-µm2 windows centered on the 40-µm-wide electrodes. The oxide is etched in 10:1 BHF for 1 min (etch rate ∼25 nm/min) for a 4% overetch. Photoresist is stripped and the wafer is cleaned, resulting in the electrode array shown Figure 1 (inset). Fluidic channels are formed by placing SU-8 molded poly(dimethylsiloxane) (PDMS) over the gold electrode substrate. Standard photolithography techniques are used to create 6-cmlong, 200-µm-wide, and 32-µm-deep channel molds with SU-8 25 (MicroChem, Newton, MA). Channels are formed by pouring PDMS (10:1 Sylgard 184 silicone elastomer base to curing agent (Dow Corning, Midland, MI) over the SU-8 mold and curing for 48 h at 37 °C. Fluidic access ports are created by punching 0.5mm holes at each end of the SU-8 defined PDMS channel. The complete cell capture microdevice is formed by bonding the PDMS channels to the patterned glass substrate (Figure 1). The PDMS substrate is removed from the SU-8 mold and cleaned in a UV ozone oven for 8 min to promote glass-PDMS irreversible bonding.23 The PDMS substrate is aligned to the glass wafer with alignment marks, contacted, and the substrate sandwich is heated at 100 °C for 15 min to promote permanent bonding. PDMA Derivitization. The bonded glass-PDMS microchannel is derivitized with poly(dimethylacrlamide) (PDMA) using a modified Hjerten coating protocol to prevent nonspecific cell adhesion to glass.24,25 First, channels are filled with 1 M NaOH for 1 h to clean and deprotonate the glass surface. After NaOH (21) Ruoslahti, E. Annu. Rev. Cell. Dev. Biol. 1996, 12, 697-715. (22) Mehrishi, J. N.; Bauer, J. Electrophoresis 2002, 23, 1984-1994. (23) Grover, W. H.; Skelley, A. M.; Liu, C. N.; Lagally, E. T.; Mathies, R. A. Sens. Actuators, B 2003, 89, 315-323. (24) Hjerten, S. J. Chromatogr. 1985, 347, 191-198.

6936 Analytical Chemistry, Vol. 77, No. 21, November 1, 2005

Figure 1. Schematic of the glass-PDMS microdevice for singlecell capture. A cell suspension enters the 200-µm-wide PDMS channel through the 0.5-mm-diameter fluidic port. Cells flow over the PDMAderivitized glass surface in the 32-µm-deep channel and are captured on the 16-µm2 exposed gold pads centered on the 40-µm-wide gold electrodes. Cells are directed to the desired electrode by applying a 50 V/cm electric field between the interdigitated electrodes (200-µm spacing). Inset: electron micrograph of an electrode showing the three exposed gold pads on the oxide-coated electrode. Bar, 30 µm.

incubation, the channel is evacuated and filled with a 0.6% (v/v) (γ-methacryloxypropyl)trimethoxysilane solution (γ, Sigma, St. Louis, MO) in 3.5 pH H2O. The bifunctional γ-solution prepares the glass surface for acrylamide polymer nucleation. During γ-solution incubation, 250 µL of dimethylacrylamide is dissolved in 4.75 mL of H2O and sparged with Ar for 1 h. After Ar sparging, 100 µL of isopropyl alcohol (IPA), 20 µL of TEMED, and 25 µL of 10% (v/v) APS are sequentially added to the acrylamide solution to form linear PDMA. The γ-solution is removed from the channel, and PDMA solution incubates in the channel for 1 h. The channel is then rinsed and dried with acetonitrile. Cell Culture. K1 and M1WT3 strains of Chinese hamster ovary (CHO) cells are cultured using standard techniques for cell capture experiments. All cell culture reagents are obtained from Gibco/Invitrogen Corp. (Carlsbad, CA) unless otherwise noted. Wild-type K1 cells (American Type Culture Collection (ATCC), Manassas, VA, CCL-61) and muscarinic receptor transfected M1WT3 cells (ATCC, CRL-1985) are cultured in parallel using identical techniques. A nitrogen-frozen stock is thawed and grown in F-12 media containing 10% (v/v) fetal bovine serum (FBS, HyClone, Logan, UT) and 1% (v/v) penicillin/streptomysin (P/S, Sigma) for 2 days at 37 °C and 5% CO2 in T-75 cell culture flasks (Corning, Acton, MA). Adherent CHO cells are grown to confluence and detached from the growth plate by adding 2 mL of trypsin/EDTA and incubating for 5 min at 37 °C. The trypsin/ EDTA is neutralized by adding 8 mL of F-12 media containing FBS and P/S. The cell suspension is centrifuged for 3 min at 5000 rpm. The supernatant is aspirated, and the cell pellet is resuspended at 1 × 106 cells/mL in media with FBS and P/S for cell capture experiments. For muscarinic activation experiments, K1 and M1WT3 cells are deprived of serum for 24 h to arrest the cells’ growth cycle. (25) McClain, M. A.; Culbertson, C. T.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 2001, 73, 5334-5338.

Figure 2. Illustration of RGD mediated cell thiolation process. (A) Integrin proteins containing a binding site for the RGD peptide motif are present on the CHO cell surface. (B) CHO cells are incubated in a 50 µM RGD-peptide solution in 1×PBS for 1 h. The CHO cell surface is effectively labeled with ∼5 × 106 thiol groups due to the presence of cysteine residues at both termini of the synthetic nine residue peptide (CCRRGDWLC). (C) Thiol functional groups on the peptide bind to exposed 16-µm2 gold pads and hold the cell in place.

Cells are grown for 2 days at 37 °C in a T-75 cell culture flasks in media containing FBS and P/S. The medium containing serum is aspirated from the flask and replaced with serum-deprived medium. The remaining cell culture steps are as previously described. RGD Binding and Thiolation. Detached CHO cells are functionalized with thiol-containing RGD peptide. The synthetic peptide, CCRRGDWLC (Sigma Genosys), is used for these experiments because the thiol-containing cysteine (C) is at the amino and carboxyl termini.26 Detached CHO cells are suspended in 5 mL of 50 µM thiol-containing RGD peptide in phosphatebuffered saline (1×PBS, pH 7.4). Cell suspension is placed in a T-25 flask and gently agitated for 1 h at room temperature to ensure RGD-cell binding and resultant cell thiolation (Figure 2). After 1 h thiolation, cell solution is centrifuged at 3500 rpm for 3 min. Supernatant containing RGD-peptide is removed and frozen for reuse. Initial RGD-cell binding optimization experiments were performed with a fluorescein-labeled synthetic peptide, FlcCCRRGDWLC. Flow cytometry with the fluorescein-labeled synthetic peptide showed that each K1 CHO cell was effectively labeled with approximately ∼5 × 106 thiol groups. CHO cells replated after thiolation show normal spreading and proliferation. CellTracker Dye Labeling. Thiolated CHO cells are suspended in 1×PBS and labeled with fluorogenic, cell permeant 7-amino-4-chloromethylcoumarin (CellTracker Blue) and 5-chloromethylfluorescein diacetate (CellTracker Green) dyes obtained from Molecular Probes (Eugene, OR). Thiolated cells are incubated in 10 mg/mL BSA and 4 µΜ CellTracker dye for 15 min to introduce tracking dye into cells. Cell suspension is centrifuged at 3500 rpm for 3 min, and then supernatant is aspirated. Cell resuspension in 1 mL of 1×PBS and centrifugation at 3500 rpm for 3 min are repeated 3 times to remove any excess dye and BSA. Cells are suspended for cell capture assay at 8.5 × 106 cells/ mL in 1×PBS using hemocytometer. Cell Capture Protocol. To prepare for CHO cell capture on microdevice, the glass-PDMS channel is primed with 1×PBS and 200 µL of cell suspension (8.5 × 106 cells/mL at room tempera(26) McMillan, R.; Meeks, B.; Bensebaa, F.; Deslandes, Y.; Sheardown, H. J. Biomed. Mater. Res. 2001, 54, 272-283.

ture) is placed in a 1-mL disposable syringe that is placed in a syringe pump (BAS, West Lafayette, IN) and connected to a microchip fluidic port through 24-gauge Teflon tubing. Cells are pumped into the channel at 50 µL/min until cell concentration in microchannel is constant. Once flow is stopped, the power supply is turned on to activate the 50 V/cm driving field between the interdigitated electrodes for a designated time (5-60 min). Unbound cells are removed by applying a constant 4 µL/min wash for 2 min. Single-cell and multicell capture events are recorded for each electrode using an epifluorescence microscope (Nikon Eclipse E800, Melville, NY) with cooled CCD camera (Media Cybernetics, San Diego, CA). After cell culture experiments, the microfluidic device is flushed vigorously with 1×PBS to remove cellular debris. The channel is aspirated and washed with IPA to clean gold pads. The microdevice is reused for cell capture experiments. Muscarinic Activation Assay. CHO cells are labeled with the cell permeant, calcium-sensitive fluo-4 AM dye, to show selective activation of the muscarinic receptor. For the muscarinic activation assay, wild-type K1 and muscarinic receptor transfected M1WT3 cells are separately cultured and labeled with the green fluorescent fluo-4 AM dye. Both cells types are labeled according to the CellTracker dye labeling protocol, except that cell media with 10% FBS and P/S is used during wash and trials instead of 1×PBS. During the 15-min BSA incubation, 6 µM cell-permeable fluo-4 AM dye is added. Excess dye is removed in the wash protocol previously described. Sequential capture of M1WT3 and K1 cells is integrated with the muscarinic activation assay using the same cell preparation and dye labeling protocol. M1WT3 cells are captured on the gold pads first to ensure cells remain viable over the course of the experiment. M1WT3 cells labeled with the CellTracker Blue and calcium-sensitive fluo-4 AM dyes are captured on odd-numbered electrodes according to the cell capture protocol. The driving field is reversed, and K1 cells labeled with the CellTracker Green and calcium-sensitive fluo-4 AM dyes are captured on even-numbered electrodes. After both cell types are captured on gold pads, a 100 µM carbachol solution in 1×PBS is added. Carbachol activates M1 receptors, present at a density of 800 fmol/mg of membrane protein in M1WT3 cells, causing the release of intracellular calcium from stores.27 Epifluorescent images are analyzed with Image-Pro software (Media Cybernetics) to determine which relative changes in calcium levels. Cell figures were processed by subtracting the background and adjusting the intensity range with ImageJ software. RESULTS AND DISCUSSION Single-Cell Capture. The cell capture microdevice was tested with both thiolated and native CHO cells, in the presence and absence of the driving electric field. For thiolated cells, a 10-min application of the electric field resulted in significant capture of single cells on positive gold electrodes from a suspension of 8.5 × 106 cells/mL (Figure 3A). Total capture increased with capture time; cells continued to be drawn to occupied gold pads leading to cell clumping at 30 min (Figure 3B). Control experiments using thiolated cells but no electric field, or a driving electric field (27) Haraguchi, K.; Rodbell, M. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 59645968.

Analytical Chemistry, Vol. 77, No. 21, November 1, 2005

6937

Figure 3. Representative experiment showing CHO cell capture on gold electrodes. (A) Thiolated K1 cells bind to the positive gold electrodes (+) when a driving 50 V/cm electric field is applied for 10 min. K1 cells are labeled with 4 µM CellTracker Green for visualization showing predominately single cell capture for short incubation times (5 single cell and 1 multicell capture events). (B) A 30-min incubation under the same conditions shows increased multicell capture events (3 single cell and 4 multicell capture events). (C) Representative sequence of five bare gold electrodes showing no capture of thiolated cells after 10 min when no field is applied. Bar, 50 µm.

without the cell thiolation, resulted in negligible capture (Figure 3C). Statistical analysis confirmed that there was a significant difference between the capture, which combined cell thiolation and electric field and the control trials lacking either component (χ2 > 45 for t g 5 min, P < 0.001). Similar to previous reports, the focal adhesion point for captured cells occurs at the edge of the gold pad.14 This suggests that the size of the pad does not limit cell capture and smaller electrodes could be fabricated for increased cell capture density. Single-cell capture efficiency was optimized by varying the duration of field application. The data were recorded based on capture events on a given electrode, each of which contains three gold binding pads. This reflects the fact that the electrode was the basic independently addressable unit. Figure 4A illustrates how single-cell capture efficiency depends on capture time. Singlecell capture attained maximal efficiency for the 10-min trials for thiolated cells in the electric field, with 63 ( 9% of rows having a single cell. Increased time led to the formation of cell clumps, which were no longer counted as single-cell capture events. Figure 4B illustrates the result of increased capture time on total capture, where any capture event on an electrode pad was included in the total. This results in a maximum total capture efficiency of 90 ( 5% for the 60-min incubation. 6938 Analytical Chemistry, Vol. 77, No. 21, November 1, 2005

Figure 4. Characterization of single-cell capture efficiency. (A) Plot of the percentage of electrode rows with single cell captured. Thiolated K1 cells in the presence of a 50 V/cm driving field have a maximum capture efficiency at 10 min of 63 ( 9% (n ) 30). Cells that are not thiolated (circle) and those thiolated but bound in the absence of an electric field (triangle) show basal levels of capture and no significant time dependence. (B) Plot of the percentage of rows with any binding event (one or more cells) versus time. Thiolated cells begin fielddriven binding in less than 5 min. The high levels of capture after 10 min are a result of multicell capture events. If either the thiol-containing peptide (circle) or the driving an electric field (triangle) are omitted, little capture is observed.

The total cell capture results exhibited an increase and then a dip before reaching a final maximum. This dip may be to be due to the manner in which the cells clumped. Capture events for the 10-min trials with thiolated cells and electric field were mostly single cells, which tended to withstand the rinse due to their lower surface area. As the cells clustered loosely, they presented a larger resistance to flow and were more likely to be removed. But as the incubation time increased further, the clumps solidified internally and in their adhesion to the surface, making them more difficult to remove with rinsing. RGD Binding and Thiolation. The use of cell surface thiolation instead of substrate modification presents several advantages. First, it relies on the robust gold-thiol bond instead

Figure 5. Sequential directed capture of two populations of CHO cells. (A) The first population of thiolated K1 cells, labeled with CellTracker Blue, is captured by applying a 50 V/cm potential to the even-numbered electrodes for 10 min. (B) A second population of thiolated K1 cells, labeled with CellTracker Green, is introduced into the channel through the opposite fluidic port and field mediated binding occurs selectively at the odd-numbered electrodes. Bar, 40 µm.

of the in situ RGD-integin interaction. Since the gold-thiol bond is 10-fold stronger than the RGD-integrin bond,28,29 this improves capture efficiency. The off-chip incubation in RGD peptide provided ∼5 × 106 thiol groups/cell, which also compares favorably to the estimated 270 000 single strands/cell with DNAmediated capture reported recently.30 Another advantage of cell labeling is that it leaves the electrodes in their native state, which is useful for sensor applications.31 In our approach, certain elements of an array of electrodes could be modified for ion selectivity, while adjacent gold electrodes are used for both cell capture and electrical measurements. In the electrode modification approach, the thiol-labeled adhesion molecules form a monolayer on the surface, potentially reducing its recording sensitivity and altering ion selectivity. The cell modification approach thus provides superior adhesion with electrical measurement flexibility. Sequential Capture. Reversing the direction of the electric field after cell capture caused the directed adhesion of cells on the opposite electrodes while leaving the previously captured cells in place. To illustrate this selectivity, the microdevice was used to capture two different populations of CHO cells on electrodes separated by 100 µm (Figure 5). Single CHO cells labeled with the cytosolic dye CellTracker Blue were first captured on the oddnumbered electrodes by setting them at ground potential with the even-numbered electrodes held at -1 V for 10 min. A single cell was captured on each of the gold pads for both electrodes pictured. The cell suspension was then rinsed from the channel and replaced with a second population of CHO cells, labeled with CellTracker Green. Once the second suspension was in the (28) Xiao, Y.; Truskey, G. A. Biophys. J. 1996, 71, 2869-2884. (29) Grandbois, M.; Beyer, M.; Rief, M.; Clausen-Schaumann, H.; Gaub, H. E. Science 1999, 283, 1727-1730. (30) Chandra, R. A.; Douglas, E. S.; Mathies, R. A.; Bertozzi, C. R.; Francis, M. B. In preparation.

Figure 6. Single-cell capture and selective activation. (A) Thiolated M1WT3 cells containing the muscarinic receptor and labeled with CellTracker Green are first captured on the even-numbered electrodes. Fluorescent cells not captured on gold pads are washed off in subsequent capture steps. (B) The thiolated K1 cells labeled with CellTracker Blue are introduced through the opposite fluidic port and are captured on the odd-numbered electrode by reversing the 50 V/cm potential. (C) Following incubation with 6 µM cell-permeable calciumsensitive fluo-4 AM dye, a solution of 100 µM carbachol in 1×PBS is introduced in the channel to selectively activate the muscarinic receptor on the M1WT3 cells. Carbachol binds to the muscarinic receptor, and the subsequent release of intracellular calcium results in an increase of green fluorescence in the M1WT3 cells, while the K1 cells exhibit only faint fluorescence. (D) The blue and green fluorescent images are overlayed, demonstrating the directed capture and selective activation of the two cell types. Bar, 40 µm.

channel, the potential was reversed. The blue fluorescent cells remained bound, while the green fluorescent cells were driven to the even-numbered electrodes by the field (Figure 5B). The final result was the patterning of single cells from two cell populations on alternating electrodes of an interdigitated array. This technique possesses the important advantage of pattern control and scalability. While it was demonstrated here for only two cell types, our approach could be used to pattern three or more groups of single cells on adjacent electrodes in a sequential manner. This suggests a new approach for parallel drug screening, in which many different cell types are monitored simultaneously in a microfluidic device.32 Alternatively, cells could be positioned adjacent to one another to study heterotypic intercellular communication at the single-cell level. Single-Cell Capture and Activation. A cell stimulation assay was integrated with sequential capture to demonstrate that the single CHO cells remained viable and active. M1WT3 cells were loaded with the intracellular calcium indicator fluo-4 AM and captured on the even-numbered electrodes (Figure 6A). K1 cells were loaded with both fluo-4 AM and CellTracker Blue and then captured in the same channel on the odd electrodes (Figure 6B). (31) Dias, A. F.; Dernick, G.; Valero, V.; Yong, M. G.; James, C. D.; Craighead, H. G.; Lindau, M. Nanotechnology 2002, 13, 285-289. (32) Gonzalez, J. E.; Oades, K.; Leychkis, Y.; Harootunian, A.; Negulescu, P. A. Drug Discovery Today 1999, 4, 431-439.

Analytical Chemistry, Vol. 77, No. 21, November 1, 2005

6939

Figure 7. Summary of assay showing the selective activation of M1WT3 cells expressing the muscarinic receptor. Single CHO cells containing calcium-sensitive fluo-4 AM dye are activated with 100 µM carbachol in 1×PBS, 10 min after electric field-enhanced capture. (A) Box plot of the percentage increase in fluorescence for the wildtype K1 cells shows no significant increase in intracellular calcium fluorescence as a result of carbachol incubation (0.3 ( 27%). Single M1WT3 cells show a significant increase of 110 ( 74% (n ) 79) after carbachol activation (t ) 12.5, P < 0.005). The central bar represents the median, the small square the mean, and the box spans the second and third quartiles. The Xs represent the extreme percentiles, and the bounding lines are the minimum and maximum values. (B) With a CHO cell activation threshold taken as three standard deviations above the K1 basal response, 5 ( 3% of the K1 cells show activation while 87 ( 4% of the M1WT3 cells show activity.

The channel was then filled with carbachol (100 µM), an acetylcholine analogue, which binds to the M1 muscarinic Gprotein coupled receptor, leading to an increase in intracellular calcium. The activation was detected by an increase in fluorescent emission by the fluo-4 AM dye in M1WT3 cells (Figure 6C) while the K1 cells remained unchanged. An overlay of the blue and green images (Figure 6D) indicates the juxtaposition of different cell types and their selective activation. The box plot in Figure 7 quantitates the on-chip activation of single cells with a statistically valid sample size. Synchronized M1WT3 cells displayed an average sustained fluorescence intensity increase of 110 ( 74% (n ) 59) 2 min after initial stimulation. Wild-type K1 cells assayed in the same manner displayed an increase of only 0.3 ( 27% (n ) 79) in fluorescent signal due to the absence of the muscarinic receptor (Figure 7A). These data were used to estimate the percentage of captured cells that remained active after capture. The mean of the K1 response plus three standard deviations was chosen as a threshold level for activation. According to this standard, 87 ( 4% of the M1WT3 cells responded to the carbachol stimulus, while only 5 ( 3% of the K1 cells met the threshold (Figure 7Β). The difference 6940 Analytical Chemistry, Vol. 77, No. 21, November 1, 2005

between the responses of the two types was statistically significant with a χ2 value of 88.3 leading to P < 0.005 for them being from the same population. Measurements on single cells display the activity of each cell within the range of the population. The assay performed here demonstrated that the cells were not only viable, but active, and capable of a measurable response to chemical stimuli after sequential capture and sustained electric field exposure. This intracellular calcium assay is a single-cell analogue to a commonly performed bulk screen for receptor agonists and antagonists of pharmacological interest. Our assay demonstrated the same gain order as the conventional fluorometric imaging plate reader format,33 while providing single-cell resolution. Because the cells remain competent for response assays, a variety of extensions of this work are possible. While cells were assayed here in the round morphology typical of adherent cells in suspension, the addition of growth media and serum causes the thiolated cells to spread out and grow under culture conditions. The capture technique could be used to direct a single cell to a microelectrode array containing several recording sites, leading to spatial resolution of the cell response or simultaneous recording of multiple analytes.34 This would provide a bioelectronic interface in which the cell would be in contact with an electrode array capable of chemical or electrical stimulation and recording. Finally, these assays could also be integrated into a microfabricated genetic analyzer for a more complete understanding of both cellular function and genetic makeup at the single-cell level. Once a cell with unique behavior is identified it could be lysed and studied by PCR to identify underlying causes or alterations.35 CONCLUSIONS The microdevice developed here presents a simple and effective means for the directed capture and analysis of single cells in a microfluidic chip. The device uses direct labeling of the cells, instead of the more common approach of tailoring the substrate for cell adhesion. The novel combination of the driving electric field and cell thiolation provides adhesion sufficient to withstand subsequent flow used for rinsing or reagent introduction. Additionally, the ability to individually direct the capture of single cells from multiple populations on neighboring electrodes in an integrated microfluidic chip presents advantages over previous work, which demonstrated multicell type patterning. Because the captured cells remain viable, chemical stimulation and optical monitoring reveal cell activity, and additional assays are made possible. This device also provides a novel platform for future single-cell genetic studies as well as the development of a bioelectronic interface for fundamental studies of cell activity. ACKNOWLEDGMENT We thank Ravi Chandra for performing the flow cytometry experiments, Jennifer Czlapinski for assistance with cell culture, (33) Hodder, P.; Mull, R.; Cassaday, J.; Berry, K.; Strulovici, B. J. Biomol. Screening 2004, 9, 417-426. (34) Marzouk, S. A. M.; Buck, R. P.; Dunlap, L. A.; Johnson, T. A.; Cascio, W. E. Anal. Biochem. 2002, 308, 52-60. (35) Lagally, E. T.; Scherer, J. R.; Blazej, R. G.; Toriello, N. M.; Diep, B. A.; Ramchandani, M.; Sensabaugh, G. F.; Riley, L. W.; Mathies, R. A. Anal. Chem. 2004, 76, 3162-3170.

and Carolyn Bertozzi for the use of her cell culture facility. E.S.D. and N.M.T. were supported by an NSF Graduate Research Fellowship and NIH Molecular Biophysics Training Grant (T32GM08295), respectively. Microfabrication was performed in the UC Berkeley Microfabrication Laboratory. This work was supported by the Chemical Sciences Division of the U.S. Depart--

ment of Energy under contract DE-AC03-76SF00098 and by the NIH under grant R01HG01399. Received for review June 10, 2005. Accepted August 19, 2005. AC051032D

Analytical Chemistry, Vol. 77, No. 21, November 1, 2005

6941